My biggest takeaway from the article is my new favorite word: cryptomorphic, meaning equivalent, but not obviously so.
The Sell ∀ ∃ as ∃ ∀ Scam
21–30 of 148 posts
Re: The Sell ∀ ∃ as ∃ ∀ Scam
#22I can walk and chew bubble gum at the same time: on one hand, yes, there's certainly a lot of Kool-Aid being drank by the AI folks. Even on HN, I constantly argue with people that genuinely think LLMs are some kind of magical black box that contain "knowledge" or "intelligence" or "meaning" when in reality, it's just a very fancy Markov chain. And on the other hand, I think that language interfaces are probably the n…
The Microsoft Research "Sparks of AGI" paper spends 154 pages describing behaviors of GPT-4 that are inconsistent with the understanding of it being a "fancy Markov chain": https://arxiv.org/abs/2303.12712
I expect that the reason people are constantly arguing with you is that your analysis does not explain some easily testable experiences, such as why GPT-4 has the ability to explain what some non-trivial and unique Python programs would output if they were run, despite GPT-4 not having access to a Python interpreter itself.
Re: The Sell ∀ ∃ as ∃ ∀ Scam
#23“Or: testing on your training data is a common way to cheat, but so is training on your test data”
The need to hold back training data for testing, and issues around testing using variants of training data versus real world data are well known.
Re: The Sell ∀ ∃ as ∃ ∀ Scam
#24Earlier quoted context omitted.
>contain "knowledge" or "intelligence" or "meaning" when in reality, it's just a very fancy Markov chain These are not mutually exclusive. If you have a Markov chain that 100% of the time outputs "A cat is an animal", then it has knowledge that a cat is an animal.
$ yes "a cat is an animal" Does `yes` also have "knowledge"?
no, but `$ yes "a cat is an animal"` does (if you know how to interpret it)
Re: The Sell ∀ ∃ as ∃ ∀ Scam
#25The author says that the builders of these systems are also the ones that run the scam. And they deliberately build it in a fashion that enables the scam. Seriously?
This is a gripe with sales and marketing. A tale as old as time.
Re: The Sell ∀ ∃ as ∃ ∀ Scam
#26This argument is disingenuous. Hyperparameter optimization is not in the same category as prompt engineering, like at all. Also no one claims like half of the things the article claims people clain.
He, imo correctly, puts them both in the category of extra degrees of freedom that can allow the user to overfit and get results that appear more impressive than the underlying reality about how the model has generalized.
But generating new prompts for GPT is incredibly cheap! You just rephrase and ask again.
That there are prompts which generate impressive results with GPT is the point. Because anyone can generate prompts - and get impressive results.
Whereas hyper parameter tuning is expensive. A system that generates good results with the right tuning doesn’t tell you much about your ability to use it to generate good results, because it will be hard to try lots of different tuning approaches to discover if you can get a useful result.
These things seem not remotely comparable.
Re: The Sell ∀ ∃ as ∃ ∀ Scam
#27This argument is disingenuous. Hyperparameter optimization is not in the same category as prompt engineering, like at all. Also no one claims like half of the things the article claims people clain.
He, imo correctly, puts them both in the category of extra degrees of freedom that can allow the user to overfit and get results that appear more impressive than the underlying reality about how the model has generalized.
Prompt engineering rather resembles the mentalists' trick of cold reading their subject, turned inside out.
Re: The Sell ∀ ∃ as ∃ ∀ Scam
#28This is a clever argument
It would be improved by a bit of additional editing in The Famous Article.
Re: The Sell ∀ ∃ as ∃ ∀ Scam
#29Are compilers a scam as well?
There exists a program every problem you have, you just have to find the code.
Re: The Sell ∀ ∃ as ∃ ∀ Scam
#30This argument is disingenuous. Hyperparameter optimization is not in the same category as prompt engineering, like at all. Also no one claims like half of the things the article claims people clain.
He, imo correctly, puts them both in the category of extra degrees of freedom that can allow the user to overfit and get results that appear more impressive than the underlying reality about how the model has generalized.
All systems are like that.